Your 20000mAh power bank only delivers 13000mAh. Fake, or just physics?
A brand customer came to me with a spreadsheet and a bad mood. His 20000mAh power bank measured 13100mAh on the bench. Buyers were leaving reviews saying fake capacity, the marketplace listing got flagged, and he wanted to know whether his cell supplier had cheated him.
I asked him one thing first. What voltage did you measure at, and what is the number on the box actually counting?
That gap — cell capacity versus output capacity — is where most fake-capacity accusations come from. Sometimes the fraud is real. Often it is not. This piece separates the two and gives you the measurements that tell them apart.
By the end you should be able to take any capacity complaint number and say in one pass: this is normal conversion loss, or this is worth chasing with your cell supplier.
The number on the box counts cells, not what comes out of the port
A lithium cell's nominal voltage is 3.7V. That is the industry-standard nominal figure for the li-ion and li-polymer cells used in power banks. The USB port does not output 3.7V though. It outputs 5V, or 9V, 12V, 15V, 20V once fast-charge negotiation kicks in. Getting from 3.7V to 5V takes a boost converter, and conversion is not free.
Count energy, not milliamp-hours. A 20000mAh pack at 3.7V holds:
Push that same 74Wh out at 5V, assuming a perfect converter:
So even at zero loss, a 20000mAh bank cannot deliver 20000mAh at 5V. The amp-hour figure shrinks because the voltage went up. Watt-hours are what holds.
Now put real conversion on top. Boost efficiency in consumer power banks generally lands between 85% and 93%, depending on circuit design, inductor choice and output current. Take 90% as a middle case:
Which is almost exactly what his bench read. 13100mAh was not fraud. It was arithmetic.
The industry distinguishes cell capacity, printed on the pack and the box, from rated output capacity, what you actually get at the port. Some makers print both. Many print one. The ones that print one are the ones getting flagged.
So when is it actually fake?
Normal conversion loss puts roughly 30% to 40% between the printed figure and measured 5V output. A gap much wider than that, or one that shows up in a specific place, is a real problem. Three we see:
Three real failure modes
- The cells are under spec. Graded, recycled or down-bin cells sold as full capacity. A pack built from cells that are really 18000mAh will read low even after you account for conversion. Genuine fraud, but the conversation is with your cell supplier, not your tester.
- The printed number is pure marketing. Nothing behind it. Common on white-label orders where the buyer specifies 20000mAh as a listing requirement and the factory prints what was asked for. A bench measurement before shipment is the only thing that catches it.
- Boost efficiency is genuinely bad. Cheap inductor, undersized MOSFET, weak thermal design. The pack holds the energy and burns a large slice of it as heat. Symptom: low output capacity and a case that gets hot under load.
Telling the first two from the third means measuring both directions. Charge empty to full and log energy in. Discharge and log energy out. If energy in is close to the cell's rated energy but energy out falls below the expected conversion loss, your boost stage is the problem. If energy in is itself far below rated, the cells are.
What the bench should actually measure
One spot-check of output mAh will not separate these. Five readings, in this order:
Five measurements
- Discharge capacity at a defined constant current. Discharge to cutoff at a fixed rate and integrate current over time. Fix the rate — capacity shifts with discharge current, so a number without a stated current is not comparable.
- Charge capacity, empty to full. Log energy in. This is your baseline for whether the cells hold rated energy.
- Conversion efficiency = energy out ÷ energy in. Compare against your design target. A design that should hit 90% and reads 78% has a boost-stage problem.
- Multi-port behaviour under simultaneous load. Two devices plugged in usually triggers a total-power limit and per-port renegotiation. Measure what each port actually gets with both loaded. A lot of "the second port is broken" complaints start here.
- Bidirectional fast charge, both directions. Fast input and fast output are separate protocol negotiations. Passing one says nothing about the other.
| What you measured | Where the problem likely is |
|---|---|
| Input capacity far below rated | Cells under spec — go to your cell supplier |
| Input fine, output low, case hot | Boost stage or thermal design |
| Output low only under multi-port load | Power allocation logic, not capacity |
| Fast input fails, output fine | Sink-side protocol negotiation |
What our unit covers
The box we supply for this is the GTI6331H-4×4, built for power banks, mobile power and storage modules. Specified scope is bidirectional power with charge/discharge cycling and capacity verification, at ±0.1% typical voltage and current accuracy, with 20+ fast-charge protocols on board — QC2.0/3.0, PD3.0/3.1, PPS, PE and the major brand protocols. It is a 4×4 configuration, four channels each direction.
Being straight about scope. Capacity verification is exactly what this box is for, and the bidirectional range is what lets you measure in and out instead of measuring one side and guessing the rest. What I will not do is invent a voltage or current range here. That lives in the datasheet, and it differs across the 6331H, 6332H and 6333H configurations in the same series. Ask for the datasheet for your specific model rather than carrying numbers over.